Preserve nonconsecutive repeated comment lines

[berkay-byte]

Closes #417.

Two long comments beginning with KEY: can wrap onto separate lines when written to a PO file. Reading that file removes the second KEY: because Message.__init__() deduplicates comments across the entire list.

Use itertools.groupby in Message.__init__() to collapse only consecutive identical comments. Repeated lines separated by other text keep their positions. Both extracted and translator comments follow this rule, including when messages are cloned. PO parsing continues to use the constructor normally.

Tests cover consecutive and separated repeats, empty comments, iterator inputs, cloning, obsolete PO entries, and wrapped-comment round trips. Existing catalog-merge deduplication remains unchanged.

Prepared with OpenAI Codex assistance.

Validation: 7,844 tests passed, 9 skipped, 2 expected failures on Python 3.14. Repository pre-commit checks passed for the changed files.

[akx]

While I don't know the rationale for the "only write unique comments" commit from nearly 20 years ago fdb3468, I think there may have been something to it.

To fix the issue in #417, namely

#. KEY:
#. foo.bar.baz...<long words more>...
#. KEY:
#. waldo.fred.plugh...<long words more>...

losing the second KEY:, I think a better patch would be to adjust Message.__init__() to keep only one copy of the same line repeated. What do you think?